Eric, Wen I worked in computers, I connected all shields to the chassis. If they want a high quality, non-resonant capacitor, I would use a planar design on the chassis wall. But then you have to observe current limiting as there is a safety issue, and as for ESD ... ! Add some ESD suppression to the planar cap.
Cortland Cortland ====================== Original Message Follows ==================== >> Date: 17-Jun-98 16:57:00 MsgID: 1061-11164 ToID: 72146,373 From: [email protected] >INTERNET:[email protected] Subj: Firewire (IEEE-1394), EMC Stories? Chrg: $0.00 Imp: Norm Sens: Std Receipt: No Parts: 1 Sender: [email protected] Received: from ruebert.ieee.org (ruebert.ieee.org [199.172.136.3]) by arl-img-4.compuserve.com (8.8.6/8.8.6/2.12) with ESMTP id TAA10647; Wed, 17 Jun 1998 19:56:54 -0400 (EDT) Received: by ruebert.ieee.org (8.8.8/8.8.8) id TAA25816 for emc-pstc-resent; Wed, 17 Jun 1998 19:50:59 -0400 (EDT) From: [email protected] X-Lotus-FromDomain: NIC To: [email protected] Message-ID: <[email protected]> List-Post: [email protected] Date: Wed, 17 Jun 1998 18:50:41 -0500 Subject: Firewire (IEEE-1394), EMC Stories? Mime-Version: 1.0 Content-type: text/plain; charset=us-ascii Sender: [email protected] Precedence: bulk Reply-To: [email protected] X-Resent-To: Multiple Recipients <[email protected]> X-Listname: emc-pstc X-Info: Help requests to [email protected] X-Info: [Un]Subscribe requests to [email protected] X-Moderator-Address: [email protected] This new technology triggers uncomfortable flashbacks to my recent experiences with USB. Both schemes float one end of the I/O cable shield to prevent ground loops, then hope for a RF ground through a capacitor (or two) to terminate the cable shield. As I lamented before, USB is rife with EMC problems stemming from the cable grounding, complex protocol/chip interface issues (who resets who?), low voltage level signalling easily corrupted by noise, and marginally tested fault-intolerant system software. (Remember the Win98 Blue Screen of Death for Gates recently? Ah, the horror.) Now, instead of a meager 12 MHz data rate that USB employed, 1394 runs at minimum 100 MHz, with multiples possible (and under development) to over 1 GHz. All this pushed though a Nintendo style connector whose shell is isolated through a capacitor. Friends. What happens when you exceed a capacitor's self resonant frequency? (Don't answer.) Hoping for some industry notes on 1394 that might offer design tips for EMC, I cruised a handful of web sites starting at the 1394 Trade Association, read page after page of info. Nothing about EMC, nothing. I surmise two possibilities: 1. The 100, 200, and 400 MHz (800 and 1000 MHz coming soon) bit rates are so wonderfully clean and weak in emissions, while being so wonderfully EMI immune, that nobody has a single complaint about it. Or, 2. It is truely problematic but nobody wants to raise the issue. (Like the classic Emperor's New Clothes story.) Does anybody have useful EMC experience to share about 1394? Can you tweak the software to exercise all the data speeds, or do you buy different speed devices and hope they all pass? (Let's count, we will need at least three for now....) Is there a transient suppressor diode that is so low in capacitance that it doesn't affect 400+ MHz signals? Ferrite beads and common-mode chokes for the signal lines that don't trash the desired signal? I welcome direct discrete emails if needed to protect the lives of the innocent. Eric Lifsey Compliance Engineer/Manager National Instruments ====================== End of Original Message =====================
